Backup systems may use incremental backups which may back up only changed data since a last backup. This may require less time, less input/output (I/O) processing, and less storage than a full backup.
In a traditional backup strategy, a backup system may perform a full backup periodically, such as once a week, and may perform incremental backups at regular intervals between full backups. In the event a system restoration is desired (e.g., after a system crash), a backup system may restore the system to a desired recovery point using the most recent full backup, and any incremental backups that occurred after the full backup.
For systems relying on incremental backups, a backup chain (i.e., the number of backups required to restore a system) may get long if a full backup is not performed regularly. Thus, it is typically inefficient to rely exclusively on incremental backups.
However, some backup systems may create a synthetic full backup (i.e., a full backup generated from portions of several incremental backups) and thus may enable the use of incremental backups without using periodic full backups (e.g., “infinite incrementals”). Users of backup systems may prefer incremental backups and thus may use “infinite incrementals” or other backup strategies which may rely on incremental backups.
However, incremental backups may increase a risk of data integrity problems. That is, if a portion of a backup is corrupted or incorrectly captured, it may not be fixed by a later synthetic backup. For example, if one or more blocks are corrupted in an incremental backup and the one or more blocks are not changed again, subsequent incremental backups may not capture a new copy of the one or more corrupted blocks. Thus, these subsequent incremental backups that reference the one or more corrupted blocks may be corrupt as well. Even a synthetic full backup may not remedy this, because the one or more corrupted blocks of the prior incremental backup may be used for the generation of the synthetic full backup. Thus, the corruption may persist for an extended period of time and may have serious implications.
Furthermore, current restoration technologies for restoring a system using incremental backups may require restoring a long backup chain (i.e., a large number of backups to comprise a full set which may be necessary to restore a system). Not only is there risk of corruption in one or more portions of the incremental backup chain, but a long backup chain may increase the number of incremental backups to be processed by restoration system or process. Processing a larger number of incremental backups may take more system memory and a longer time.
In view of the foregoing, it may be understood that there may be significant problems and shortcomings associated with current system backup technologies.